DiffSinger
An open-source singing voice synthesis system that controls pitch, energy and breathiness and supports deployment in OpenUTAU.
DiffSinger generates synthetic singing voices for voice developers and music production. OpenVPI maintains this fork of MoonInTheRiver/DiffSinger, with acoustic models and adjustable vocal expression controls.
- Vocal expression: Predict and adjust pitch, energy and breathiness to shape how a voice sings.
- Singing audio generation: Produce singing audio with acoustic models and accelerated diffusion sampling.
- OpenUTAU deployment: Use the system's deployment support to make voices available in OpenUTAU.
- Dataset preparation: Use companion MakeDiffSinger pipelines and tools to create datasets for voice development.
The system uses Python and is open source under Apache 2.0. Its main tutorial and resource hub is in Chinese. The project prohibits using its functionality to generate someone's voice without their consent.
Logotyper
